i oecd cincome inequality in oecd countries: data and explanations a b atkinson cesifo working paper...
TRANSCRIPT
INCOME INEQUALITY IN OECD COUNTRIES:DATA AND EXPLANATIONS
A B ATKINSON
CESIFO WORKING PAPER NO. 881CATEGORY 10: EMPIRICAL AND THEORETICAL METHODS
FEBRUARY 2003
PRESENTED AT CESIFO CONFERENCE ON GLOBALIZATION,INEQUALITY AND WELL-BEING, NOVEMBER 2002
An electronic version of the paper may be downloaded• from the SSRN website: www.SSRN.com• from the CESifo website: www.CESifo.de
CESifo Working Paper No. 881
INCOME INEQUALITY IN OECD COUNTRIES: DATA AND EXPLANATIONS
Abstract
There is much disagreement about both the facts and the explanations of income inequality.Even if we confine attention to OECD countries, we find people arguing that there has been agreat U-turn, with inequality rising sharply after its post war fall, and others who believe thatthe speed of change is glacial. In order to evaluate the historical record, we need data for along run of years. The present paper reviews evidence about covering the period 1945-2001for nine OECD countries. It is widely believed that rising inequality is attributable totechnological change and to globalisation. The second part of the paper argues that these areonly part of a complex story. Household incomes depend on public policy and on sources ofincome apart from work. What is happening at the top of the distribution may need to beexplained quite differently.
JEL Code: H0, E6.
Anthony B AtkinsonDepartment of Economics
Nuffield CollegeOxford OX1 1NFUnited Kingdom
1
CESifoRevised
Income Inequality in OECD Countries: Data and Explanations1
Tony Atkinson
Introduction
The debate about globalisation and world inequality has several
dimensions, and these are often confused. Some people are concerned with
absolute measures of well being and others with relative inequality. Sometimes it
is inequality between countries that is the driving force; at other times the key
variable appears to be inequality within countries. Changes in income inequality
may be attributed to internal structural dynamics (the “Kuznets curve”) or to the
interdependence of countries (“international division of labour”). It is therefore
not surprising that participants in the debate find themselves engaged in
controversy and that outside observers find themselves puzzled. This paper deals
with just one part of the puzzle: the relative distribution of income within OECD
countries. But, in concentrating on inequality within rich countries, I am not
leaving controversy behind. There are strong differences of view – both about the
facts and about their explanation. These two aspects form the content of the two
main sections of the paper.
Section A is concerned with the empirical evidence about income
inequality. Many people are firmly convinced that the period of diminishing
inequality after the Second World War has come to an end and been replaced by a
period of widening differences. Harrison and Bluestone (1988) have christened it
1 Revised version of paper prepared for the CESifo conference on “Globalization,Inequality and Well-Being” in Munich, November 8-9, 2002. I am grateful to theparticipants at the conference for their most helpful comments, which have led to
2
the “great U-turn”. According to Alderson and Nielsen, “after four decades of
moderating inequality, income inequality in the United States began to increase
around 1970. Since then it has risen at a steady rate” (2002, page 1246). They go
on to suggest that the upswing has an “international character”. Cornia and Court
describe how “the Golden Age, a period of stable global economic growth
between the 1950s and early-mid 1970s, witnessed declines in income inequality
in a number of countries (with some exceptions). This trend was reversed over the
last two decades as country after country has experienced an upsurge in income
inequality” (2001, page 7). In contrast to the “great U-turn” hypothesis stands the
belief that there has been little distributional change within countries, supported
by a number of studies of the variation of inequality over time. Gustaffson and
Johansson find for 16 industrialised countries that “the correlation between the
Gini coefficient and the time-variable is almost zero” and that there is only “a
weak U-shaped relationship” (1999, page 591). Melchior, Telle and Wiig (2000)
conclude that in industrialised and high income developing countries inequality
has not on average changed much between 1960 and 1990. On this view, the pace
of change is glacial.
There is therefore disagreement about what is to be explained. A quarter
century ago, Pen commented, “the great debate on income distribution goes on and
on; and not even the facts are beyond dispute. In this age of microdata, computers
and Theil coefficients, a straightforward question like: “Has inequality in the
developed countries of the West diminished over the last decades?” provokes
different answers from different observers” (Pen, 1979, page 682). Today we can
say the same about the question as to whether inequality has increased.
significant revisions, and to Professor Gert Wagner for his supplying the most
3
Section B is concerned with some of the possible explanations. Here again
there are different views. In the field of income distribution, there are many forces
at work, and these interact, as is well illustrated by two frequently discussed
elements: globalization and the welfare state. In what has become a widely
accepted explanation for rising income inequality, these elements have been
invoked to explain how a common force (globalization) can have differing effects
in different countries depending on the extent of their welfare states.
Globalization and/or technological change have led to a shift in demand away
from unskilled labour. The distributional consequences of such a shift depend on
the form of the welfare state. Where wages are freely flexible, then the shift in
demand causes increased wage dispersion. Where the welfare state provides an
effective floor to wages, then the result is unemployment of unskilled workers.
Thus, we have a unified explanation of what is happening in the United
States (increased wage premium for skilled relative to unskilled workers) and in
Continental Europe (high unemployment). This story is not however fully
convincing. As I have argued (Atkinson, 1999), this “Transatlantic Consensus” is
open to question. Even on its own terms it is incomplete, and when we step
outside the textbook model the conclusions can be rather different. In Section B, I
consider the supply response in terms of investment in skills, and the link between
factor prices (wages) and the distribution of personal incomes. This both reveals
alternative explanations, involving the capital market as well as the labour market,
and helps explain differences across countries. Moreover, the conventional theory
focuses on unskilled workers and the bottom of the income distribution. In a
number of OECD countries, however, it is increased inequality at the top that is
recent results from the German Socio-Economic Panel.
4
most evident. In the final part of Section B, I investigate the long-run experience
of top income groups in four OECD countries and point to the need for alternative
explanations.
A The Empirical Picture
Issues of Definition
Inequality of what? Among whom? Measured how? Here my preferred
definition is the distribution of disposable annual income among households with
total household income being adjusted for household size and/or composition and
each household being counted as many times as there are individuals. This choice
is open to question.2 Money income is only a partial measure of social welfare,
and expenditure may be preferred as a measure of standard of living. Inequality of
consumption may have risen less than that of income (Krueger and Perri, 2002).
On the other hand, it may be argued that our concern should be with potential
rather than actual use of resources. The study of poverty in OECD countries (see
Atkinson, 1998) has distinguished between approaches concerned with the
standard of living and approaches concerned with the right to a minimum level of
resources.3 The difference is neatly encapsulated by the fact that the US poverty
line used to be lower for women than for men, a difference which was justified in
terms of the difference in minimum food requirements, but which clearly could
not be justified on a minimum rights basis. One can interpret the move to a
2 These definitional issues are discussed in, among other places, Atkinson,Rainwater and Smeeding (1995), Atkinson and Brandolini (2001), and the ExpertGroup on Household Income Statistics (2001).3 This distinction is relevant to debate about the World Bank’s $1 a day poverty line,where many commentators assume that standard of living is the sole concern, whereascritics of the World Bank approach may favour a rights-based concept.
5
uniform scale for men and women as an implicit change in the underlying
concept. The revealed preference of governments, via the medium of their
statistical agencies, is indeed for the distribution of income, and it is on this
definition that I concentrate here.
Differences in definition may affect not only the level of measured
inequality but also the change over time. This should be emphasised since there is
a tendency to assume that differences in definition can be treated as a fixed effect.
Three examples serve to illustrate this point and to underline the limitations of the
concept used in the empirical results presented here. The first is the use of annual
rather than permanent income. To the extent that annual income exhibits
transitory variation that can be offset by borrowing and lending, it overstates
inequality of permanent income. It is possible that the economic changes of
recent decades have been associated with increased volatility (see Gottschalk and
Smeeding, 2000, Section 5, for a review of the limited evidence), in which case
the rise in annual income inequality may overstate the rise in permanent income
inequality. This would not be captured by a time constant dummy variable. A
second example is that the analysis makes no allowance for within household
inequality. In the UK, poverty measured on a family unit basis has typically been
significantly higher than poverty measured on a broader household basis. It has
recently become clear (Webb, 2002) that the difference has declined over time.
Poverty among families has increased substantially in the UK but by less than on
a household basis. The trend in the two series is different. The third example is
the shortfall of measured income from a comprehensive definition. Important
sources are typically missing from official statistics, and these change over time.
Capital gains, stock options, and remuneration offshore became more important in
6
the latter part of the Twentieth Century, leading inequality to be increasingly
under-stated. Conversely, tax reductions in a number of OECD countries in the past
two decades may have led to a re-arrangement of income, so that more income
appears in taxable form. In the US it is argued that there has been a shift from the
corporate to the personal tax base (Gordon and Slemrod, 2000). Moreover, countries
differ in the degree of comprehensiveness of their income definition, and this will
affect the degree to which the changes over time can be compared.
In summarising the extent of inequality in Section A, I make use of the Gini
coefficient, which has a graphical interpretation as the ratio of the area between the
Lorenz curve and the diagonal to the whole triangle, but which is only one
particular way of reducing the whole distribution to a single number. It is
perfectly possible for the distribution to change significantly but for the Gini
coefficient to remain unchanged. There could be redistributive forces working in
different directions at different points. Demographic change could be causing a
rise in the proportion of pensioners with relatively low incomes, while at the same
time progressive taxation is reducing the shares of top income groups. This is one
reason why, in Section B, I turn to looking at the shares of top income groups,
which could be moving in a different pattern from the Gini coefficient.
Data
My time frame is the second half of the twentieth century. I believe that
the recent inequality movements can only be understood in the light of the
historical experience. Indeed, while attention is confined here to the period 1945-
2000, an even longer perspective would be desirable. I also attach considerable
weight to high frequency series. Reference to “high frequency” may sound absurd
7
when the term is more conventionally applied to weekly or daily financial data.
What I am suggesting here is that where the underlying data are collected annually,
then we should use the full set. In particular, I have serious doubts about the practice
adopted recently by the OECD (for example; Förster, 2000) of taking single
observations for “mid-1970s”, “mid-1980s” and “mid-1990s”. Such a procedure can
misrepresent the dynamic pattern. A single year can be highly misleading.
I have therefore assembled income distribution data for nine OECD
countries:4 Canada, the UK and the US, Italy, the Netherlands, and West
Germany, Finland, Norway and Sweden. As the order of listing indicates, they
may be seen as belonging to three groupings: Anglo-Saxon, mainland European,
and Nordic countries. I should stress that these data are not comparable across
countries. The exercise is different from that conducted by the Luxembourg
Income Study, which has very valuably put income distribution data on a broadly
comparable basis. Their estimates are however limited to a selection of years. My
quest for annual series means that I have to use the estimates produced in the most
part independently by official statistical agencies. They reflect particular national
features. For example, the official US estimates of the Gini coefficient shown in
Figure 1 relate to the distribution of income before tax. In this respect they differ
from most other countries.
4 This compilation draws heavily on work with Andrea Brandolini of the Bank ofItaly. Brandolini (1998) has data for the pre-war period as well. Among earliersources with time series covering a range of countries are Atkinson (1997),Atkinson, Rainwater and Smeeding (1995, Chapter 5), Flora et al (1987), Fritzell(1993), Gardiner (1997), Gottschalk and Smeeding (1997 and 2000), Jäntti andDanziger (2000), Kraus (1981), Kuznets (1963), Morrisson (2000), Roberti(1974), Sawyer (1976), and Stark (1977).
8
The Hypotheses under Examination
Earlier I contrasted the “great U-turn” view with that of “glacial change”.
What would constitute evidence in support or contradiction of these hypotheses?
To begin with, what would represent distributional change sufficient to be “non-
glacial”? This is usually considered in terms of statistical significance. For
example, Statistics Canada (2002, page 25) suggests, with a sample of some
30,000 households, that a change of 1 percentage point should be considered
statistically significant. But what about the economic significance? To get some
idea, suppose that the tax and transfer system were approximately linear, as with a
uniform tax credit and a constant tax rate. Then, if government spending absorbs
20 percent of tax revenue, a redistributive tax of 5 percentage points would reduce
a market Gini of 48 percent by 3 percentage points.5 Raising the tax rate from 20
percent to 25 percent would be a major political shift, so this suggests that a
change of 3 or more percentage points can certainly be taken as economically
significant. In what follows I take such a 3-percentage point yardstick, while
recognising that it is essentially arbitrary. The mental experiment does however
underline another aspect of the test implicitly being applied: that we are
concerned with a sustained change, not with year-to-year transitory variation. A
series of Gini coefficients that went from 25 percent to 30 percent and back over
the business cycle may exhibit non-glacial behaviour but be essentially trend-free.
Evidence in favour of the view that there had been non-glacial change would be
that there is a period of a number of years when the level of inequality was 3
5 A gross income of Y becomes a net income of (1-t)Y+A, where A is the value ofthe tax credit. Since A is the same for everyone (with appropriate equivalisation),the Gini is (1-t) times the value for gross income divided by the mean net incomerelative to the mean gross income, which is assumed to be 0.8.
9
percentage points or more different from a period of a similar number of years in
the past.
The existence of a significant rise in recent years does not imply that the
U-hypothesis is validated. The change could be a step function. Or the time path
could look like that of an aircraft take-off: horizontal for a period and then a steep
ascent. In examining the case for the U-hypothesis, I shall be asking whether or
not there has been a period of significant decline (again I apply the 3-percentage
point criterion) followed by a period of significant increase. This is a different
criterion from the more usual practice of fitting a quadratic equation to explain the
Gini coefficient (G) as a function of time (T): G = a – bT + cT2. One can then test
whether the coefficients on time and time squared are significant (and negative
and positive respectively). Such an approach is, in my view, too restrictive. The
first reason is that it imposes a particular structure on the U. It does not allow for
situations where the decline and rise in inequality are separated by a period of
stability: a flat-bottomed U. Most importantly, the quadratic implies that the rise
in inequality is continuing. It assumes a U without a serif. In contrast, a U with a
serif indicates that the rise has come to an end. Given that we are especially
concerned with the recent past, I add a third hypothesis to those being examined:
that inequality is continuing to trend upwards in the 1990s (the continuing rise
hypothesis).
The second reason why I do not simply fit a quadratic is that, as the
diagrams below seek to highlight, we do not have a single consistent series for
each country. There are discontinuities that cannot be simply modelled by a
dummy variable. There are different series telling different stories. The evidence
is of variable quality. These considerations are not readily incorporated into a
10
purely econometric analysis. One has to apply judgment as well as statistical
packages.
Finally, I should note that the hypotheses are being considered here in
relation to the Gini coefficient, which is only one summary measure of the
distributional change. Other measures may yield different results. This could go
either way. As noted earlier, stability of the Gini coefficient may conceal
significant changes in, say, the share of the bottom 20 percent. We may therefore
be incorrectly rejecting the hypothesis of non-glacial change. The converse may
also be true. A redistribution from the top 10 percent to the next 40 percent may
reduce the Gini coefficient significantly, but a test based on the share of the bottom
50 percent would indicate no significant change.
The Pictures
Figure 1 shows that we can indeed assemble data covering many years, in
this case for the US. The data are rich, but they need to be used with care. For
instance, while the US Current Population Survey provides data from the 1940s,
the earlier form of the official series referred only to “families”, leaving out
“unrelated individuals” (i.e. people living on their own). Rather than use this
partial population, I have selected the “unofficial” estimates of Budd (1970) for
the earlier years, together with the OBE synthetic series, drawing on survey
information, tax data and the national accounts. In the US, as in other countries,
the picture is a patchwork rather than a single seamless garment. Moreover, even
what may appear to be a single series often exhibits discontinuities, as there are
changes in definitions, in methods of data collection, or in data processing. In
assembling the data series I have tried to identify and label the most important
11
breaks in the series. One notable break in the CPS series was that in 1993 when
the data collection changed from paper and pencil to computer assisted
interviewing, and when there was a large increase in the top codes (that for
earnings rose from $299,999 to $999,999). This was important, since there was a
large rise in recorded inequality in that year, and estimates (Weinberg, 1996,
footnote 3) indicate that these changes could account for one half of the recorded
increase.
So what support do we find for the various hypotheses? Certainly there
has been a rise post-1968. By 1992 the Gini coefficient had risen by some 5-
percentage points – well in excess of my 3-percentage point yardstick. The non-
glacial change hypothesis is accepted. What about the U-hypothesis? Visual
inspection indicates an episode of falling inequality from 1961 to 1968, when the
Gini fell by 2½ percentage points. This could be called the first part of the U, but
falls short of the 3-percentage point yardstick, and it was preceded by a period
that shows virtually no trend. To quote Lindert, “for the US, the shift to more
equal pre-fisc incomes lasted only a quarter century, from 1919 to 1953. ... Then
it stopped altogether” (Lindert, 2000, page 195). (The U might be clearer for the
post-fisc distribution.) The Kennedy-Johnson years apart, the US post-war pattern
looks more like an aircraft take-off: horizontal for a period and then a steep
ascent. The key question concerns when it levels off. Here the break in 1993
makes it difficult to draw firm conclusions about the continuing change
hypothesis. The fact that the coefficient in 2001 was 1 percentage point higher
than in 1993 (allowing for the new population controls and expanded sample)
suggests that cruising altitude may not have been reached, but the next few years
will be watched with interest.
12
For the UK, we can see from Figure 2 that the picture looks even more
like a patchwork. Evidence from household surveys effectively begins in 1961.
For the earlier years, we have a synthetic series, based on income tax and other
data, that suggests, like the US, little overall change up to the 1960s. Then the
pattern departs from that in the US in that there appears to have been a distinct fall
of 3 percentage points in the 1970s. The fall was more than reversed in the 1980s.
Two features stand out about the UK experience. The first is the sheer magnitude
of the rise from 1984 to 1990: the Gini coefficient in the UK rose by more than 1
percentage point a year. The U is lop-sided to the right. Overall there was an
increase in the Gini coefficient of 10 percentage points. Any theory must explain
why the UK was twice as severely affected as the US. Even if part of the rise was
reversing the fall in the 1970s, the 1990 figure was 6.7 points higher than the
highest value recorded in the 1960s. The second feature is that the 1990s did not
show a continuing upward trend. Allowing for the break in the series, the 2000
Gini coefficient is the same as that for 1990. So we have a serif U.
The UK evidence in Figure 2 demonstrates one characteristic of recent
changes emphasised by the OECD: that most countries have seen a rise in
inequality of market income (income before taxes and transfers). If we are
interested in the impact of economic forces such as increased international
competition then it is presumably market incomes that interest us most. The same
story of rising market income inequality is exhibited by the Canadian data shown
in Figure 3: since 1981 the Gini has risen by more than 5 percentage points. But
the inequality of disposable income remained unchanged up to 1996. Until that
point, the tax and transfer system appears to have been successful in offsetting
any exogenous forces making for greater inequality. (I stress “exogenous” since
13
market income itself is potentially affected by the existence of taxes and
transfers.) What does the Canadian evidence tells us about the U-hypothesis? A
supporter may discern a U-shape, but the magnitude of variation is very limited:
the range in 30 years is from 35 percent to 37.8 per cent. On this basis, Canada is
on the margins of glacial change.
Turning to Scandinavia, we find that Sweden is an example of the
problems of data comparability. At face value, there is (Figure 4) a clear U-
pattern. But as far as the downward movement is concerned, a lot rests on the
observation for 1967, of which Gustafsson and Uusitalo say that “because of
some differences between the two data sets the comparability is less satisfactory”
(1990, page 84). Gustafsson and Johansson (1999) for this reason exclude the
1967 observation from their analysis of changes in inequality over time and the
official SCB series starts in 1975. From 1975 to 1981 the fall was smaller but still
sufficient (3 percentage points) to qualify as non-glacial change. The rise from
1981 to 1990 is 2.8 percentage points, so up to this juncture the U is muted. In
considering the 1990s, there are definitional issues. The official “headline” series,
unlike in most other European countries, includes realised capital gains, which
leads the series to be much more volatile in the 1990s. This causes differences in
the picture depending on the definition chosen. From 1993 to 1999, the rise is 3.7
percentage points if realised capital gains are included, but 1.6 percentage points
if they are excluded (Statistics Sweden, 2001, page 3). The judgment reached
regarding the continuing rise hypothesis depends therefore on the concept of
income employed.
A contrast is often drawn between Norway and Sweden. The picture
(Figure 5) in Norway is however far from clear. Indeed those commentating on
14
the Norwegian experience have reached different conclusions. According to
Bojer, the period 1970-1984 in Norway showed “great stability in the distribution
of personal income” (1987, page 257). According to Ringen, the distribution
from 1970-1986 “has not been stable” with first a rise and then a fall in inequality
(1991, pages 6 and 7). So that, while there has been a clear rise in the 1990s, by
3.3 percentage points from 1990 to 1997, it is less clear from the existing
evidence that this is the second part of a U. On the other hand, Finland (Figure 6)
provides much clearer evidence of a U, even allowing for the break in the series.
The Gini coefficient for disposable income for 1966 was 31.8 in 1966, fell to
around 20, and then rose again to 26.6 percent in 2000. The increase from 1993 to
2000 is over 5 percentage points, providing strong support for the continuing rise
hypothesis.
In Continental Europe, we find a marked decline from 1959 to 1977 in the
Netherlands (Figure 7). As described by van Zanden, “the CBS [Central Bureau of
Statistics] figures show quite a marked fall in inequality from 1962 onwards, a fall
which continues into the first half of the seventies. About ten years of stability
followed, after which a slight increase in inequality can be registered, starting in
1983. ... Thus the long-term fall in income inequality which had run through the 20th
century seems to have come to an end half way through the 1980s” (van Zanden,
1998, page 177). The rise from 1983 to 1990 was 3 percentage points, so that the
Netherlands just qualifies as a U-shape. It is however a U with a serif, since the
1990s indicate no continuation of the rise: the Gini coefficients for 1990 and 1997
are identical. Moreover, it should be noted that the data for years before 1977
relate to tax units rather than households. In West Germany (Figure 8), the earlier
period is surrounded by uncertainty. Both the budget survey (EVS) based
15
estimates and the DIW synthetic estimates show falls in the Gini coefficient of
more than 3 percentage points, but the timing is quite different. The EVS
estimates show a fall from 1962 to 1973, but this is not mirrored in the DIW
synthetic estimates, which show a rise. The preference today is to use household
survey data, but it is not evident that we should simply believe one series and not
the other. The EVS is based on a quota sample, and a lot of weight attaches to the
first observation in reaching the conclusion that inequality fell significantly. The
DIW estimates incorporate information from other sources, notably the tax
returns. In the more recent period, the EVS and the Panel survey show a similar
upward trend, but the overall rise from 1988 to 2001 is less than 3 percentage
points. The U is less than clearcut. The rise over the 1990s can only be described
as “modest” (less than 2 points). Finally, the Italian picture has to be compiled
from different series, with different income definitions, reflecting changes in the
underlying sources (Figure 9). They certainly suggest significant change, but is
not a simple U. The time path of the Gini coefficient in its homeland over this
period is more like a W. As summarised by Brandolini, “from the early 1970s
until 1982 ... the inequality of household incomes fell dramatically. In the mid-
1980s, it showed some tendency to grow; a further decline in 1989-91 was soon
reversed, and in 1995 the Gini coefficient was back to the value of 1980” (1999,
Temi version, page 51). The W appears to have a serif, since the Gini remained
fairly flat from 1993 to 2000.
Conclusions
What do I conclude regarding the three hypotheses of non-glacial change,
U-shape, and continuing rise? As I have stressed, I do not believe that
16
unequivocal answers can be given, as the data at our disposal are a patchwork that
requires interpretation. (Moreover, conclusions may have to be revised as new
data become available.) The significance of observed changes is a matter for
judgment. These qualifications should be borne in mind when considering the
summary in Table 1, where bold entries are in accord with the hypothesis, normal
type entries do not support the hypothesis, and italic entries are unclear or “on the
margin”. Applying the 3 percentage point criterion to the Gini coefficient, there
is general support for the view that income inequality has exhibited non-glacial
change, with the possible exception of Canada. What about the U-hypothesis? In
three of the nine cases, a U-shape can clearly be discerned (UK, Finland and
Netherlands), to which we can add the US if the Kennedy-Johnson years are
treated as a significant decline, and the more muted cases of Canada and Sweden.
The existing evidence in Norway and West Germany is unclear and Italy is more
like a W. Where recent rises in inequality have followed earlier episodes of
falling inequality, there may be a U or a U: the rise may have levelled off or be
continuing. The changes over the 1990s divide into three: 3 countries where the
Gini coefficient increased by more than 3 points (Norway, Finland and Italy), 3
countries where there was a modest rise (US, Canada and West Germany), 3
countries where there was no overall change (UK and Netherlands) or different
series tell different stories (Sweden).
The principal conclusion that I draw is that there was a wide diversity of
experience, even among OECD countries. This means that the explanations must
be sufficiently rich to allow a variety of outcomes.
17
B Seeking Explanations
Reduced Demand for Unskilled Workers
Diversity of outcome is well illustrated by the consensus view about the
cause of rising inequality. The textbook explanation is that there has been a shift
in demand away from unskilled towards skilled labour, as a result of increased
competition from newly industrialising countries (NIC) as a result of globalisation
or of technical change biased towards skilled labour, or both these factors
operating in conjunction. The reduced demand for less skilled labour means that,
with relative supplies of the two kinds of worker fixed in the short-run, in a free
labour market there will be a rise in the premium for skilled workers and a decline
in the relative wage of unskilled workers. This is taken to be the explanation of
rising wage dispersion in the US. In Continental Europe, unemployment benefits
and/or minimum wages place a floor and mean that the relative wage of the
unskilled cannot fall. According to this view, the demand shift explains the higher
unemployment in Europe. There is a unified explanation as to how a single cause
has a differential impact on the US and on Continental Europe.
For all its attractions, the standard textbook explanation has important
limitations. As the more sophisticated international trade literature has recognised
(Davis, 1998), we cannot look at two parallel universes with 2 trading regions (in
one case US and NIC, and in the other Europe and NIC). We need a minimum of
3 trading regions (US, NIC and Europe). If in a unified analysis the US and
Europe both produce the good that faces NIC competition, then the wage floor in
Europe determines the relative goods prices (in a standard two good two-factor
Heckscher-Ohlin model, where it is assumed that one good uses unskilled labour
relatively intensively). If the minimum wage is unchanged, this prevents the
18
relative price from falling. The US is therefore unaffected by increased trade.
Europe bears the brunt in terms of unemployment. We have one arm of the
“Transatlantic consensus” explanation, but not the other arm. As Neary has
noted, the assumption of fixed relative wages in this highly simplified model
“imposes an implausible degree of structure on the world economy” (2001, note
3). The result is better stated as a tendency: European unemployment tends to
prop up American wages. On this basis, the outcry about globalisation should be
in Europe not in the US. On the other hand, Europe may be protected to a degree
because it has already become specialised in goods that use skilled labour
intensively. If, in the standard trade theory model, we allow the EU to have
become specialised, then the consequences of opening trade are different (see
Oslington, 1998). Suppose that the NIC is specialised in the other good (which
uses unskilled labour intensively). Their entry into world trade drives down its
price and hence causes the relative wage of unskilled workers to fall in the US,
which produces both goods. We then have the US side of the textbook
conclusion, but not the European side, since the demand for the good using skilled
labour intensively is higher, causing a fall in unemployment. Either way, one or
other part of the "textbook" view cannot apply.
The trade explanation is not therefore a simple application of standard
theory of the Heckscher-Ohlin variety. The model has to be expanded. Nor is this
the only limitation. Here I consider two further aspects that apply when considering
both trade and technology versions of the demand shift theory: (a) supply side
reactions and alternative theories of pay determination, and (b) the relation between
factor prices and personal incomes.
19
The Supply Side
Suppose that we now consider the relation between the skill premium and
the supply of skilled labour. In the simplest form of human capital theory,
acquisition of a skill requires S years of training, during which the person is paid
nothing, so that the earnings career is postponed by S years. To provide the same
present value of earnings over a fixed working life, the earnings of skilled
workers have to be higher by a factor erS where r is the real interest rate. If there
are no barriers to acquiring skills, then this factor determines the long-run
equilibrium wage differential. It should be noted that the existence of this
differential does not imply inequality: it is a compensating differential equalising
net advantage. If the labour market were initially in equilibrium, then a rise in the
skill premium, caused by increased relative demand for skilled workers, will
induce an increase in the supply of skilled workers. A rise in the premium for
college educated workers means that more people choose to stay on at college.
At this point we should note that we are assuming away barriers to entry
into skilled employment. One such barrier is the capital market. People may not
be able to borrow to finance their education, and their parents may lack the
necessary capital. Families may be trapped in a low level equilibrium (Atkinson,
1997). This process may have been exacerbated by the fact that real interest rates
are now higher than they were in the 1970s (when they were negative). Borrowing
has become more costly. To the extent that this is the case, we have arrived at a
quite different explanation of increased earnings dispersion: that there has been an
increase in the factor erS. The explanation is to be found not in the labour market,
on which economists tend to focus, but in the capital market.
20
Even where the barriers can be overcome, the acquisition of human capital
takes time. The stock of college-educated labour cannot adjust at once. But in the
case of the US, we are talking about a rise in earnings dispersion that began more
than a generation ago. Surely by now the supply would be increased? To quote
James Galbraith, “nearly thirty years since the start of rising inequality, many
millions have acquired the skills appropriate to the age. Word-processing,
accounting and calculating on spreadsheets, e-mail and the Internet, computer
graphics and publication, computer-aided design: none of this is any longer
esoteric. Yet the readjustment of incomes to a wider and more equal distribution
of skill levels hasn’t even begun to happen” (1998, pages 7 and 8). Absence of
adjustment may be due to the nature of the “skills” that are rewarded in the labour
market. “Increasing economic value now attaches to individual attributes of a
kind less likely to be achieved through the educational system than ascribed
through processes of socialisation within generally more advantaged families and
communities” (Erikson and Goldthorpe, 2002, page 40). In his analysis of
telephone operators, Bresnahan concluded that: "workers need more people skills,
and less cognitive ones" (1999, page F405). It may therefore be that, in seeking an
explanation, we are focusing on the wrong kinds of skills.
Alternatively, labour markets may function differently. In his account of a
period of narrowing wage differentials, Reder states that "the long-run decline in
the skill margin in advanced countries has not occurred slowly and steadily.
Instead, the skill margin appears to have remained constant for relatively long
periods of time and then to have declined sharply within a very few years" (1962,
p.408). Institutional labour economists have typically assumed that supply and
demand only place limits on the possible wage differentials, with other factors such
21
as bargaining or social convention determining where between these limits wages
actually lie. Social conventions may play an instrumental role in removing the
indeterminacy. Social codes may however enter more directly into economic
behaviour. Akerlof (1980) describes a model where individual utility depends not
only on income but also on reputation that is based on conformity with the social
code. The loss of reputation if one departs from the social code depends on the
proportion who believe in the code, which is undermined if people cease to observe
it.
Such a reputational approach is applied in Atkinson (1999) to the relation
between wages and productivity. Suppose that there is a social code, or pay norm,
that limits the extent to which individual earnings increase with earnings potential.
Where this code is followed, people are paid a fraction (less than unity) of their
productivity plus a uniform amount. Such a policy involves a degree of
redistribution and low productivity workers can be expected to subscribe to the pay
norm. But other workers will also accept it, even where they could be paid more if
they broke the norm, since - if they believe in the norm - by breaking it they would
suffer a loss of reputation. The extent of the loss rises with the proportion of the
population who at that time believe in the norm. Employers are also concerned with
their reputations. When they create a job, it is determined in advance whether or not
it is paid according to the pay norm. The profitability of the job depends not only on
the pay but also on the acceptance of the job by the worker with which it is matched.
Matching is assumed to follow a random process, but is only successful where
employer and worker either both observe the code or both do not. Employers
determine their pay policy (i.e. whether or not to observe the social code) on the
basis of comparing expected profitability, which depends on the proportion, and
22
characteristics, of workers who accept different pay offers. The expected
profitability of breaking the social code has to exceed the cost of the consequential
loss of reputation, which is assumed to vary across employers, so that some
employers may observe the code while others depart from it. There will therefore be
a proportion of jobs which accord with the pay norm. If the proportion of the
population who believe in the pay norm is less (greater) than this, then the extent of
belief is assumed to grow (fall).
There is therefore a dynamic process of adjustment. As Akerlof has
shown, the process is likely to be of the "tipping" kind identified by Schelling
(1978). Depending on the initial conditions, a society converges to a high level of
conformity with the social code, or to the virtual absence of conformity. In this
kind of situation, an exogenous shock may switch the society from an equilibrium
with conformity to the pay norm, and hence relatively low wage differentials, to
an equilibrium where everyone is paid on the basis of their productivity. Such an
exogenous shock may have been a fall in the weight attached by employers to
reputation. Or, reflecting again changes in the capital market, it may be that
greater weight is attached to short-run profits. We may therefore observe a
discrete change in the wage distribution, an episode of increasing dispersion,
which is not reversed when the original cause is removed. In this way we have
arrived at a more textured explanation of rising wage dispersion, with rather
different implications from the textbook supply and demand account. It can, for
example, explain why the same shock may have different effects on different
countries, depending on the extent of differences in underlying productivity.
23
From Factor Prices to Personal Incomes
The conventional explanation begins and ends with the relative wages of
skilled and unskilled workers. However there are several steps between relative
factor prices – the central focus of economics textbooks – and the distribution of
disposable income among households, which is what we have been examining in
the empirical section. These steps are typically ignored, but are potentially
important.
Suppose that a man loses his job as an unskilled steelworker, as a result of
competition from imports, and has to take a lower paid job in the retail sector. The
impact on the household disposable income cannot be predicted solely from this
information, nor can we place him in the distribution without knowing about:
• His other sources of income;
• The income of other members of the household;
• The size and composition of the household in which he lives;
• The consequential changes in taxes and transfers.
Working from the bottom, we can see that the net of tax change in income is
reduced by the existence of progressive taxation. At an overall scale, with a
marginal tax rate of 30 percent, a linear tax system means that a 5 percentage
point rise in the Gini coefficient for market incomes translates into a 3½
percentage point rise in the Gini for disposable income. On this basis, we would
expect to find differences across countries in the impact on disposable income of
a given shock to market income according to the degree of progression of their tax
systems.
This might lead us to expect a larger rise in inequality of disposable
income in the US and the UK, where tax systems are less progressive. But we
24
have also to remember that these countries have in-work benefits that generate
high implicit marginal rates of tax (and mean that the tax system is not linear): the
Earned Income Tax Credit in the US and the Working Families Tax Credit in the
UK. These benefits are tapered with income, and – for much of the relevant range
– increase the marginal tax rate. The worker facing a pay cut would find in the
UK that his tax credit rises to offset part of the fall in gross earnings. The implicit
marginal tax rate can be 70 percent or higher. Tax benefit policy therefore
considerably moderates the impact of increased wage dispersion.
Equally, we have implicitly been assuming that the low skilled worker is
at the bottom of the distribution. But his or her ranking depends on the income of
the other members of the household and on the size and composition of the
household. It is a very different situation if he is the sole earner in a household
with 4 children from that where he is a single man. Moreover, we cannot assume
that all male unskilled workers are married to unskilled workers. If, at the other
extreme, every unskilled worker is married to a skilled worker (and there are
equal numbers, all married), then a change in the skill premium does not affect the
household distribution of income. There will be a change in the distribution
within the household, and this may well be important, but it is not captured in the
usual statistics.
Lastly, there are other sources of market income besides earnings, notably
capital income. He may be like the soldiers who, in Tawney’s phrase, went off to
the First World War carrying the sum total of their possessions. But today it
seems more likely that he has significant assets, both real property and financial
assets, and liabilities. The liabilities may exceed the assets for some people, but
the household sector as a whole has substantially positive net worth. Capital
25
income is not limited to a minority capitalist class. It is therefore surprising that
capital is entirely missing from the textbook account of rising inequality. The
production function should surely include capital as well as skilled and unskilled
labour? Globalisation, and other forces influencing the skill premium, is likely to
have affected the rate of return to capital. The search for shareholder value may
have meant that the steelworker’s job disappeared, but the higher capital income
may be funding his early retirement.
For these reasons, a rise in inequality of individual gross earnings may be
considerably moderated when we follow through the implications for the
distribution of household equivalised disposable income.
Top Incomes
The textbook story concerns the plight of the unskilled worker in a
globalising world of rapid technological change. It tells us less about the impact
of these forces on upper income groups. In the same way, the Gini coefficient is a
convenient summary indicator, but may not be a good guide to what is happening
at the top of the distribution. If we focus just on the shares of top income groups,
then we can make use of longer-run data derived from income tax returns. Tax
data have many shortcomings. They relate to gross (pre-tax) income. They are
affected by tax evasion and avoidance. The definitions of income and the income
unit follow those of the income tax legislation, which varies both over time and
across countries. Capital income is recorded to differing degrees in different
countries, and the same applies to executive compensation in kind and in stock
options. Income tax data do however have the merits of covering the whole
second half of the twentieth century and of providing considerable detail on the
26
top of the distribution. Figures 10 and 11 show the results of studies for four
OECD countries. To the US, UK and Canada, I now add France, as it was the
research of Piketty (2001) for France that revived use of this source – which had
of course been extensively used half a century ago by Kuznets (1953) for the US.
In order to relate the tax data to the whole population, we need reference
totals for the number of tax units and total income, in order to calculate the share
in total income of the top x percent of total tax units. The total of tax units is
relatively easy to approximate from population statistics, but the studies cited
have adopted different approaches to the income totals, particularly with regard to
the inclusion or exclusion of transfer incomes. In view of this, I have concentrated
here on the distribution within the top income groups: the share of the top 1
percent in the total income of the top 10 percent (Figure 10) and the share of the
top 0.1 percent in the total income of the top 1 percent (Figure 11). This is in itself
of considerable interest. In presenting income distribution data in terms of decile
shares, we often overlook the fact that income within the top 10 percent is highly
unequally distributed. In the same way, the top 1 percent cannot be regarded as a
homogenous class. Pen (1971, page 9) comments on the widespread tendency to
overstate the number of the very rich. People tend to assume that most members
of the top 1 percent are CEOs or pop singers or footballers.
Figures 10 and 11 demonstrate the inequality within the top income
groups. If one takes the UK income distribution in 1979 (a year of relatively low
inequality), then according to the income tax data (the UK is shown by squares in
Figures 10 and 11) the share of the top 10 percent in gross income was 25.3
percent: i.e. 2½ times their proportionate share. Within the top 10 per cent, the
top 10 percent (i.e. the overall top 1 percent) had a share of 20.9 percent of the
27
total income of the decile group: i.e. twice their proportionate share. And the top
10 percent within the top 1 percent (the overall top 0.1 percent) had a share of 22
percent of the total income of the percentile group. The similarity of these
numbers reflects the fact that the upper tail of the distribution has approximately a
Pareto form, and in this sense they are not surprising. They serve nevertheless to
highlight the way in which – at whatever point one slices the upper reaches of the
distribution – the remaining section is characterised by a similar degree of relative
inequality. Assuming that the cumulative distribution F within the top group is
such that (1-F) is proportional to y-α, where y is income, then the within-group
share of the top 1 percent within the top 10 percent, denoted by S1/S10 is (0.1)(1-
1/α), which falls with the Pareto exponent, α. This method of estimating the Pareto
coefficient was proposed by Macgregor (1936), who noted that it made a bridge
between Pareto and Lorenz. For this reason, to draw a distinction from other
methods of estimating the Pareto coefficient, I refer to it as the Pareto-Lorenz
coefficient.
Looking across the four countries, we see two striking features of the post
war period in Figures 10 and 11. (The US data are shown by circles; the Canadian
data by diamonds; and the French data by a solid line.) The first is the similarity
in the downward trend in the relative shares for the first twenty-five years. The
implied Pareto exponents were already quite similar, but they rose. In 1949 the
Pareto-Lorenz coefficients derived from Figure 10 are 2.24 in Canada and France,
1.96 in the US and 1.81 in the UK. By 1970 the Pareto-Lorenz coefficients had
risen to 2.50 in France, 2.54 in the US, 2.57 in the UK and 2.67 in Canada. For
the shares of the top 0.1 percent within the top 1 percent (Figure 11), the series
were already close and the relative shares fell together.
28
The second striking feature is that the latter part of the century saw a sharp
rise in the relative shares and fall in the Pareto-Lorenz coefficient in the Anglo-
Saxon countries, but not in France. We have a much clearer U shape, without
apparent serifs. By 1998, the implied Pareto-Lorenz coefficients from Figure 10
had fallen to 2.10 in Canada, 1.95 in the UK and 1.83 in the US. The US and the
UK had switched position, but the values are essentially the same as in 1949. It is
true that the bottom of the U came later in the UK and Canada (around 1978), and
that we have to allow for a break in comparability in the UK series with the
introduction of independent taxation in 1990, but the overall time-path in the three
Anglo-Saxon countries is similar. In contrast, as pointed out by Piketty (2001),
France did not see a sharp upward movement in inequality. Compared with 1970,
the French Pareto-Lorenz coefficient in 1998 is fractionally down (Figure 11) or
actually higher (Figure 10).
Possible Explanations of Top Income Inequality
The evidence just presented for the top of the distribution provides clear
support, in three Anglo-Saxon countries, for the U hypothesis and for the
continuing rise hypothesis. The upward arm of the U may be explained by the
same forces of globalisation and technology. According to the "superstar" theory
of Rosen (1981), the expansion of scale associated with globalisation and with
increased communication opportunities has raised the rents of those with the very
highest abilities. Where the second-best performer’s market share is limited by the
“reach” of the top performer, and this reach is extended by technical changes such
as those in ICT, and by the removal of trade barriers, then the earnings gradient
becomes steeper. The explanations for earnings inequality at the top may however
29
be different; we may have to consider capital income rather then earned income.
We have to cast the net wider if we are seeking to explain the evolution of top
income shares.
Here I confine myself to noting that the empirical representation in terms
of the Pareto exponent provides a direct link to a number of theories advanced in
the past (and largely ignored today). If the upper tail can be approximated by the
Pareto distribution, then the share of the top 0.1 percent within the top 1 percent is
a linear function of (1/α):
Log[S1/S10] = b (1/α - 1) (1)
In the case of earnings, one set of theories that lead directly to predictions
concerning the Pareto exponent are those dealing with executive remuneration in
a hierarchical structure. The model advanced by Simon (1957) and Lydall (1959
and 1968, page 129) leads to an approximately Pareto tail to the earnings
distribution, where 1/α = loge[1+ increment with promotion] divided by loge[span
of managerial control]. As noted by Phelps Brown (1977, page 309), plausible
values for the span and increment imply values of α higher than observed in the
actual distribution. On the other hand, the theory suggests one approach to
understanding the variation in α: that the size of the increment for promotion fell
over the first half of the post war period and then rose, except in France.
The differing experience across countries in the second period could be
explained by applying a “tipping” model similar to that described above. Those at
the top of a “flat” hierarchy may, conforming with the pay norm, accept lower pay
than they could secure in a firm with a “steep” hierarchy (with generous stock
options). Their adherence may however be a function of the extent of compliance
in the economy as a whole, both because this affects the potential loss of
30
reputation from breaking the code and because it affects the chance of being
better paid if they entered the market for another job. The existence of executive
remuneration surveys provides in this case a powerful instrument for assessing the
degree of conformity, as does the presence of external directors on remuneration
committees. In this situation, an interior equilibrium may be unstable, with the
economy tending to one extreme or the other. With a flat hierarchy, the pay
increments are the reservation wage; with a steep hierarchy they are constrained by
what shareholders tolerate. What we have observed in the US, UK and Canada may
have been a progressive move from one equilibrium to another. It is possible for
example that a reduction in firm time horizons in some countries has led to such a
shift.
What can explain the first part of the period, when α was rising? Here we
have to recognise that earnings are only part of the income of the top groups. As
Piketty (2001) has demonstrated for France, the composition of income changes
radically within the top 10 percent. For those in the “first” 5 percent, earnings (in
1998 in France) accounted for 90 percent of their income; for those in the top 0.01
percent, capital income accounted for over 60 percent of total income. This brings
us to theories concerned with the accumulation of capital. Meade (1964)
developed a model of individual wealth holding, allowing for accumulation and
transmission of wealth via inheritance, and this model has been analysed in a
general equilibrium setting by Stiglitz (1969). With equal division of estates at
death, a linear savings process, and persistent differences in earnings, in the long-
run the distribution of wealth mirrors the distribution of earnings (Atkinson and
Harrison, 1978, page 211). In contrast, alternative assumptions about bequests can
generate long-run equilibria where there is inequality of wealth even where
31
earnings are equal. Stiglitz shows how the operation of primogeniture in passing
on wealth can lead to a stable distribution with a Pareto upper tail, with
1/α = loge[1 + sr(1-t)] / loge[1+n] (2)
where sr(1-t) is the rate of accumulation out of wealth, r being the rate of return
and t the tax rate, and n is the rate of population growth (Atkinson and Harrison,
1978, page 213). For stability (and α greater than 1), the overall population
growth rate has to exceed the rate of accumulation by the wealthy. The model is
highly stylised but again provides a direct explanation for the decline and then
rise of the top shares over the post war period. In the first post war decades, the
net rate of return was increasingly reduced by progressive taxation and by
inflation; the 1980s and 1990s then saw a recovery of the real rate of return and –
in the Anglo-Saxon countries (although to varying degrees) – reductions in top
income tax rates. These models treat wealth accumulation as deterministic.
Champernowne (1936, 1973) set out a rich theory of the evolution of incomes
subject to stochastic shocks. This has been developed by Vaughan (1978, see also
1979), who shows that the Pareto exponent is a function of the degree of
volatility. This provides a second element that can contribute to the explanation of
the U shape of top income shares, if volatility first fell over the post war period, as
the Golden Age became established, and then increased in the more turbulent later
years of the century.
These models indicate some promising routes to the specification of
econometric equations linking the share of the top 1 percent within the top 10
percent (and the share of the top 0.1 percent within the top 1 percent) to macro-
economic and other variables, but they need to be developed further. Given the
32
significant changes at the top of the distribution, we need to look more widely
when seeking explanations.
Prospects for the Future
Those drawing attention to rising income inequality are right to be
concerned about the future. With the possible exception of Canada, there has been
significant change in the distribution of income in all nine OECD countries
examined here. In some countries, such as the UK, the increase in recorded
inequality is considerably larger than the rise in the US that has attracted so much
attention. At the same time, these recent changes should not be extrapolated into
an inexorable rise in inequality in the future. There are at least three reasons for
not drawing such a conclusion. First, the empirical evidence suggests that in some
countries the rise does not appear to be continuing: we have a U rather than a sans
serif U. Secondly, the distribution of income is a highly complex phenomenon,
and it would be surprising if a single explanation sufficed for all countries and all
periods. Globalisation and ICT may together have reduced the job opportunities
of those with low levels of skill, but there are also other forces in operation and
these forces may change direction. For example, I have suggested that part of the
explanation of greater inequality, especially at the top of the distribution, is the
rise in the net rate of return to capital. If the rise is unwound in the next decade,
with investors enjoying lower yields than in the 1980s and 1990s, we may expect
the increased inequality to be reversed. Thirdly, we must not lose sight of the role
of policy. I have concentrated on the explanation of variations in inequality of
market income, but changes in tax and social transfer policy have played a major
role in increasing inequality in a number of countries.
33
Table 1 Summary of Empirical Evidence on Gini Coefficient
Country Non-Glacial Change? U-Turn? Continuing Rise?
United States Significant change:Gini in 1992 higher by 5points than in 1968.
Not clear in postwarperiod. Yes, if treatKennedy-Johnson yearsas significant decline.
Modest rise from 1993to 2001.
UnitedKingdom
Significant change:Gini in 1990 higher by 10points than in 1978.
Yes, U. No overall change inGini 1990 to 2000.
Canada On the margin. Muted. Modest rise inGini from 1990 to1999.
Sweden Significant change. Muted up to 1990. Different pictures fromdifferent series.
Norway Significant rise in 1990s. Picture unclear. 3 point rise from 1990to 1997.
Finland Significant fall and rise. Yes, U. 5 point rise 1993 to2000.
Netherlands Significant change. Yes, U. No overall change inGini 1990-1997.
West Germany Significant change: 19734.4 points lower than1962.
Picture unclear. Modest rise 1990 to2001.
Italy Significant falls and rises. More like W. Sharp rise 1990-1993,stable 1993-2000.
34
Appendix
Figure 1 US Income Inequality 1945-2000
38
39
40
41
42
43
44
45
46
47
1945
1947
1949
1951
1953
1955
1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
Gin
i %
CPS survey
OBE synthetic estimates
B
EBreak
Estimates of Budd
C
35
Figure 2 UK Income Inequality 1949-2000
20
25
30
35
40
45
50
55
1945
1947
1949
1951
1953
1955
1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
Blue Book before tax income
Blue Book after tax income
IFS equivalised disposable income
ET market income
ET disposable income
ET equivalised market income
ET equivalised disposableincome
L
KJ
I
G
B
HF
C
D
36
Figure 3 Canada Income Inequality 1965-1999
34
36
38
40
42
44
46
48
50
52
1945
1947
1949
1951
1953
1955
1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
Gin
i coe
ffici
ent % SC market income
SC disposable incomeK
E B
DSC gross income
Wolfson market income
37
Figure 4 Sweden Income Inequality 1967-1997
15
20
25
30
35
40
45
50
55
60
1945
1947
1949
1951
1953
1955
1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
Gin
i coe
ffici
ent %
SCB equivalised market income
Break
G+U equivalised disposable income
SCB without capital gains and household basis
SCB equivalised disposable income
Break?
KJ
D
F
38
Figure 5 Norway Income Inequality 1970-1997
20
25
30
35
40
45
1945
1947
1949
1951
1953
1955
1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
Gin
i coe
ffici
ent %
Ringen equivalent market income
Ringen equivalent disposable income
SSB equivalised disposable income
Bojer disposable income
K
J
HBreak
Break
E
B
39
Figure 6 Finland Income Inequality 1966-2000
15
20
25
30
35
40
45
50
1945
1947
1949
1951
1953
1955
1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
Gin
i coe
ffici
ent %
Equivalised market Income
Equivalised disposable income
Equivalised gross income
Break
Break
Break
40
Figure 7 Netherlands Income Inequality 1947-1997
20
25
30
35
40
45
50
1945
1947
1949
1951
1953
1955
1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
%
20
25
30
35
40
45
50
Gross income from tax data
CBS equivalised disposable income
CBS disposable income
Tax units Households
C
D
E
B
41
Figure 8 West Germany Income Inequality 1950-2001
20
25
30
35
40
45
1945
1947
1949
1951
1953
1955
1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
Gin
i coe
ffici
ent % DIW disposable income synthetic series
EVS equivalised disposable income
Panel equivalised disposable income
EVS equivalised market income
Panel equivalised market income
B
C
D
F
E
42
Figure 9 Italy Income Inequality 1968-2000
27
29
31
33
35
37
39
41
43
1944 1946 1948 1950 1952 1954 1956 1958 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
Gin
i coe
ffici
ent %
Person weights equivalised disposable income
Household weights disposable Income
BC
D
E
FExcluding imputed rent, interest and dividends
Excluding interest and dividends
Total income
Figu
re 1
0 Sh
are
of to
p 1%
in to
p 10
%
202224262830323436
1945
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
% income of top 10%
Brea
k in
UK
US FR
UK
CA
Figu
re 1
1 Sh
are
of to
p 0.
1% in
top
1%
2025303540
1945
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
% income of top 1%
Brea
k in
UK
US
US
FR
FR
UK
UK
CA
CA
45
Data Sources
Figure 1:B Brandolini, 1998, Table A1, col 1a.C Budd, 1970, page 255.E U.S. Census Bureau, 2002, Table A-2.
Figure 2:B Royal Commission on the Distribution of Income and Wealth(RCDIW), 1979, page 165.C Atkinson and Micklewright, 1992, Table BI.1.F 1961 to 1974 from RCDIW, 1977, page 247; 1975, 1977, 1979,1981, 1983, 1985-1987 from Office for National Statistics, Economic Trends(ET), May 1990, page 118; 1976, 1978 and 1980 from ET, January 1982, page105; 1982 from ET, November 1983, page 87; 1984 from ET, July 1986, page103.G and J 1977 from ET, March 1997, page 53; 1978 to 1993 from ET, May2002, page 72.H and K ET, May 2002, page 72.I 1961 to 1974 from RCDIW, 1977, page 249; 1980 from ET,January 1982, page 100; otherwise as F.L 1961-1991 from Goodman and Webb, 1994, page A2; 1992-1999/2000 from Clark and Goodman, 2001, Figure 2.
Figure 3:B, D and K 1971-1979 from Statistics Canada, 1996, page 34; 1980-1999 fromStatistics Canada, 2001, Table 705.E Wolfson, 1986, page 348.
Figure 4D and F Up to 1995 from Statistics Sweden, 1999, page 3; 1996-1999 fromStatistics Sweden, 2001, page 3.J Gustafsson and Uusitalo, 1990, page 85.K 1991 to 1995 Eriksson and Pettersson, 2000, page 173; 1996-1999from Statistics Sweden, 2001, page 3.
Figure 5:B Income Distribution Survey,http://www.ssb.no/english/subjects/05/01/incdist/main.htmlE Bojer, 1987, page 251; 1973, 1976 and 1979 from Bye andHøyland, 1981.H Ringen, 1991, page 6.J Ringen, 1991, page 9.
Figure 6:Uusitalo, 2001, page 4; 1987 onwards from Statistics Finland website:http://tilastokeskus.fi/tk/el/tulo/gini.html.
46
Figure 7:B Brandolini, 1998, Table A9.C supplied by Central Bureau of Statistics, June 1999.D and E Trimp, 1996, page 32.
Figure 8:B 1950-1968 from DIW, 1973, page 224; 1970-3 from DIW, 1974,page 312; 1975 onwards from Guger, 1989, Chart 1.C and D Hauser and Becker, 2001, page 89.E and F SOEP results supplied by G Wagner, Tabelle 1b (see Becker et al,2003, but note that the results here apply an equivalence scale equal to the squareroot of household size).
Figure 9:B and C supplied by Bank of Italy.D, E and F Brandolini, 1999, page 227.
Figure 10 and 11:Canada: Saez and Veall, 2002, Table B1.France: Piketty, 2001, pages 620-1.UK: sources described in Atkinson, 2002.US: Piketty and Saez, 2001, Table A1.
47
References
Akerlof, G A, 1980, "A Theory of Social Custom, of which Unemployment may beone Consequence", Quarterly Journal of Economics, vol 95: 749-775.
Alderson, A S and Nielsen, F, 2002, “Globalization and the Great U-Turn: IncomeInequality Trends in 16 OECD Countries”, American Journal of Sociology, vol107:1244-1299.
Atkinson, A B, 1997, "Bringing Income Distribution in from the Cold", EconomicJournal, vol 107: 297-321.
Atkinson, A B, 1998, Poverty in Europe, Basil Blackwell, Oxford.
Atkinson, A B, 1999, Is Rising Income Inequality Inevitable?, WIDER AnnualLecture 3.
Atkinson, A B, 2002, “Top Incomes in the United Kingdom over the TwentiethCentury”, Discussion Paper in Economic and Social History, No 43, University ofOxford.
Atkinson, A B, and Brandolini, A, 2001, “Promise and Pitfalls in the Use of"Secondary" Data-Sets: Income Inequality in OECD Countries”, Journal ofEconomic Literature, vol 34: 771-799.
Atkinson, A B and Harrison, A J, 1978, Distribution of Personal Wealth in Britain,Cambridge, Cambridge University Press.
Atkinson, A B and Micklewright, J, 1992, Economic Transformation in EasternEurope and the Distribution of Income, Cambridge, Cambridge University Press.
Atkinson, A B, Rainwater, L and Smeeding, T M, 1995, Income Distribution inOECD Countries, Evidence from the Luxembourg Income Study, Paris, OECD.
Becker, I, Frick, J R, Grabka, M M, Hauser, R, Krause, P, and Wagner, G G,2003, “A Comparison of the Main Household Income Surveys for Germany: EVSand SOEP” in R Hauser and I Becker, editors, Reporting on Income Distributionand Poverty, Springer Verlag, Berlin.
Bojer, H, 1987, “Personlig inntektsfordeling i Norge 1970-1984, Tidsskrift forSammfunnsforskning, vol 28: 247-258.
Brandolini, A, 1998, “A Bird’s-Eye View of Long-Run Changes in IncomeInequality”, Temi di discussione, Banca d'Italia.
Brandolini, A, 1999, “The Distribution of Personal Income in Post-War Italy:Source Description, Data Quality, and the Time Pattern of Income Inequality”,Giornale degli Economisti e Annali di Economia, vol. 58: 183-239, and Temi didiscussione Number 350, Banca d’Italia.
48
Bresnahan, T F, 1999, "Computerisation and wage dispersion", Economic Journal,vol 109: F390-F415.
Budd, E C, 1970, “Postwar Changes in the Size Distribution of Income in the U.S.”,American Economic Review Papers and Proceedings, 1970, vol 60: 247-60.
Bye, T and Høyland, T E, 1981, Inntektsbegreper – inntektsfordeling, Rapport81/31, Statistisk Sentralbyrå (Central Bureau of Statistics of Norway).
Champernowne, D G, 1973 (originally written in 1936), The Distribution of Incomebetween Persons, Cambridge University Press, Cambridge.
Clark, T and Goodman, A, 2001, “Living Standards under Labour”, IFS electionbriefing note number 4, Institute for Fiscal Studies, London.
Cornia, G A and Court, J, 2001, “Inequality, Growth and Poverty in the Era ofLiberalization and Globalization”, UNU/WIDER Policy Brief No 4, WIDER,Helsinki.
Davis, D R, 1998, “Does European Unemployment Prop Up American Wages?National Labor Markets and Global Trade”, American Economic Review, vol 88:478-494.
DIW (Deutsche Institut für Wirtschaftsforschung), 1973, “Einkommensverteilungund –schichtung der privaten Haushalte in der Bundesrepublik Deutschland 1950bis 1970”, Wochenbericht, No 25, Berlin.
DIW (Deutsche Institut für Wirtschaftsforschung), 1974, “Das Einkommensozialer Gruppen in der Bundesrepublik Deutschland im Jahre 1973”,Wochenbericht, No 35, Berlin.
Erikson, R and Goldthorpe, J H, 2002, “Intergenerational Inequality: A SociologicalPerspective”, Journal of Economic Perspectives, vol 16, no 3: 31-44.
Eriksson, I and Pettersson, T, 2000, “Income Distribution and Income Mobility-Recent Trends in Sweden” in R Hauser and I Becker, editors, The PersonalDistribution of Income in an International Perspective, Springer, Berlin, pages158-175.
Expert Group on Household Income Statistics (The Canberra Group), 2001, FinalReport and Recommendations, Ottawa.
Flora, P, Kraus, F and Pfenning, W, 1987, State, Economy and Society in WesternEurope 1815-1975, Volume II, Frankfurt, Campus Verlag.
Förster, M, 2000, “Trends and Driving Factors in Income Distribution and Povertyin the OECD Area”, Labour Market and Social Policy Occasional Papers 42.
Fritzell, J, 1993, “Income Inequality Trends in the 1980s: A Five-CountryComparison”, Acta Sociologica, vol 36: 47-62.
49
Galbraith, J K, 1998, Created Unequal, The Free Press, New York.
Gardiner, K, 1997, “A survey of income inequality over the last twenty years – howdoes the UK compare?” in P Gottschalk, B Gustafsson and E Palmer, editors,Changing Patterns in the Distribution of Economic Welfare, Cambridge UniversityPress, Cambridge, 36-59.
Goodman, A and Webb, S, 1994, For Richer, For Poorer, Institute for FiscalStudies, Commentary No 42.
Gordon, R and Slemrod, J B, 2000, "Are 'Real' Responses to Taxes Simply IncomeShifting Between Corporate and Personal Tax Bases?", in J B Slemrod, editor, DoesAtlas Shrug?, Harvard University Press, Cambridge.
Gottschalk, P and Smeeding, T M, 1997, “Cross-National Comparisons of Earningsand Income Inequality”, Journal of Economic Literature, vol 35: 633-87.
Gottschalk, P and Smeeding, T M, 2000, “Empirical Evidence on Income Inequalityin Industrial Countries”, in A B Atkinson and F Bourguignon, editors, Handbook ofIncome Distribution, Elsevier, Amsterdam.
Guger, A, 1989, “The Distribution of Household Income in Germany”, WIFOWorking Papers, no 35, Vienna.
Gustafsson, B, Aaberge, R, Cappelen, A, Pedersen, P J, Smith, N and Uusitalo, H,1999, “The distribution of income in the Nordic countries: changes and causes” inM Kautto, M Heikkilä, B Hvinden, S Marklund, and N Ploug, editors, NordicSocial Policy, Routledge, London.
Gustafsson, B and Johansson, M, 1999, “In Search of Smoking Guns: What MakesIncome Inequality Vary Over Time in Different Countries?”, American SociologicalReview, vol 64: 585-605.
Gustafsson, B and Uusitalo, H, 1990, “Income Distribution and Redistributionduring Two Decades: Experiences from Finland and Sweden” in I Persson, editor,Growing Equality in the Welfare State, Oxford University Press, Oxford.
Harrison, B, and Bluestone, B, 1988, The Great U-Turn, Basic Books, New York.
Hauser, R and Becker, I, 2001, Einkommensverteilung im Querschnitt und imZeitverlauf 1973-1998, Bundesministerium für Arbeit und Sozialordnung, Bonn.
Jäntti, M and Danziger, S, 2000, “Income Poverty in Advanced Countries”, in A BAtkinson and F Bourguignon, editors, Handbook of Income Distribution, Elsevier,Amsterdam.
Kraus, F, 1981, “The Historical Development of Income Inequality in WesternEurope and the United States”, in P. Flora and A. J. Heidenheimer (eds.), The
50
Development of Welfare States in Europe and America, Transaction Books, NewBrunswick, N.J.
Krueger, D and Perri, F, 2002, “Does Income Inequality Lead to ConsumptionInequality?”, NBER Working Paper No 9202.
Kuznets, S, 1953, Shares of Upper Income Groups in Income and Savings, NationalBureau of Economic Research, New York.
Kuznets, S, 1963, “Quantitative Aspects of the Economic Growth of Nations: VIII.Distribution of Income by Size”, Economic Development and Cultural Change, vol11, No. 2, part II: 1-80.
Lindert, P H, 2000, “Three Centuries of Inequality in Britain and America”, in A BAtkinson and F Bourguignon, editors, Handbook of Income Distribution, Elsevier,Amsterdam.
Lydall, H F, 1959, “The Distribution of Employment Incomes”, Econometrica,vol 27: 110-115.
Lydall, H F, 1968, The Structure of Earnings, Clarendon Press, Oxford.
Macgregor, D H, 1936, “Pareto’s Law”, Economic Journal, vol 46: 80-87.
Meade, J E, 1964, Efficiency, Equality and the Ownership of Property, Allen andUnwin, London.
Melchior, A, Telle, K and Wiig, H, 2000, “Globalisation and Inequality”, RoyalNorwegian Ministry of Foreign Affairs, Studies on Foreign Policy Issues, Report6B: 2000, Oslo.
Morrisson, C, 2000, “Historical Perspectives on Income Distribution: The Case ofEurope” in A B Atkinson and F Bourguignon, editors, Handbook of IncomeDistribution, Elsevier, Amsterdam.
Neary, J P, 2001, “Competition, Trade and Wages”, Research Paper 2001/03,Globalisation and Labour Markets Programme, University of Nottingham.
Office for National Statistics, various years, "The Effects Of Taxes And Benefits OnHousehold Income", Economic Trends.
Oslington, P, 1998, “Factor market linkages in a global economy”, DiscussionPaper, University of Oxford.
Pen, J, 1971, Income Distribution, Allen Lane, London.
Pen, J, 1979, "A Clear Case of Leveling: Income Equalization in the Netherlands",Social Research, vol 46: 682-694.
Phelps Brown, H, 1977, The Inequality of Pay, Oxford University Press, Oxford.
51
Piketty, T, 2001, Les hauts revenus en France au 20ème siècle, Grasset, Paris.
Piketty, T and Saez, E, 2001, “Income Inequality in the United States, 1913-1998”,NBER Discussion Paper forthcoming.
Reder, M W, 1962, "WAGES: Structure". International Encyclopedia of the SocialSciences, Macmillan, New York, vol 16: 403-414.
Ringen, S, 1991, “Households, Standard of Living, and Inequality”, Review ofIncome and Wealth, vol 37: 1-13.
Roberti, P, 1974, “Income Distribution: A Time-Series and a Cross-Section Study”,Economic Journal, vol 84: 629-38.
Rosen, S, 1981, "The Economics of Superstars", American Economic Review, vol71: 845-858.
Royal Commission on the Distribution of Income and Wealth, 1977, Report 5. ThirdReport on the Standing Reference, London, HMSO.
Royal Commission on the Distribution of Income and Wealth, 1979, Report 7.Fourth Report on the Standing Reference, HMSO, London.
Saez, E and Veall, M, 2002, “The Evolution of High Incomes in Canada, 1920-2000”, discussion paper.
Sawyer, M, 1976, “Income Distribution in OECD Countries”, OECD EconomicOutlook, Occasional Studies, Paris, OECD.
Schelling, T C, 1978, Micromotives and Macrobehavior, Norton, New York.
Simon, H, 1957, “The Compensation of Executives”, Sociometry, vol 20: 32-35.
Statistics Canada, 1996, Income after tax, distributions by size in Canada, 1994,Statistics Canada, Ottawa.
Statistics Canada, 2001, Income Trends in Canada 1980-1999, CD-Rom13F0022XCB, Statistics Canada, Ottawa.
Statistics Canada, 2002, Income Trends in Canada 1980-1999 User’s Guide,Income Statistics Division, Statistics Canada, Ottawa.
Statistics Norway, 1993, Inntekts- og Formuesstatistikk 1982, 1984-1990,(Income and Property Statistics 1982, 1984-1990), Oslo
Statistics Sweden, 1999, “Income distribution survey in 1997”, Statistiskameddelanden, If 21 SM 9901, Stockholm.
52
Statistics Sweden, 2001, “Income distribution survey in 1999”, Statistiskameddelanden, If 21 SM 0101, Stockholm.
Stark, T, 1977, The Distribution of Income in Eight Countries, Royal Commissionon the Distribution of Income and Wealth, Background Paper, No 4, London,HMSO.
Stiglitz, J E, 1969, “Distribution of income and wealth among individuals”,Econometrica, vol 37: 382-397.
Trimp, L, 1996, “Inkomens 1959-1994”, Sociaal-Economische Maandstatistiek, vol13, December: 31-34.
U.S. Census Bureau, 2002, Money Income in the United States: 2001, CurrentPopulation Reports, Series P-60, No. 218, Washington, D.C., U.S. GovernmentPrinting Office.
Uusitalo, H, 2001, “Changes in Income Distribution during a Deep Recession andAfter”, Central Pension Security Institute, Helsinki.
van Zanden, J L, 1998, “The Egalitarian Revolution of the 20th Century” in LSoltow and J L van Zanden, editors, Income and Wealth Inequality in theNetherlands 16th-20th Century, Het Spinhuis, Amsterdam, 175-191.
Vaughan, R N, 1978, “Savings behaviour and the distribution of wealth”, paperpresented at the European Meeting of the Econometric Society, Geneva.
Vaughan, R N, 1979, “Class Behaviour and the Distribution of Wealth”, Reviewof Economic Studies, vol 46: 447-465.
Webb, J, 2002, “Always With Us? The Evolution of Poverty in Britain 1886-2002”,thesis presented for D Phil, Nuffield College, Oxford.
Weinberg, D H, 1996, “A Brief Look at Postwar U.S. Income Inequality”, CurrentPopulation Reports, P60-191, U.S. Census Bureau, Washington, D.C.
Wolfson, M C, 1986, “Stasis Amid Change – Income Inequality in Canada 1965-1983”, Review of Income and Wealth, vol 32: 337-69.
CESifo Working Paper Series(for full list see www.cesifo.de)
________________________________________________________________________
815 Rainer Niemann and Dirk Simons, Costs, Benefits, and Tax-induced Distortions ofStock Option Plans, December 2002
816 Jan-Egbert Sturm and Barry Williams, Deregulation, Entry of Foreign Banks and BankEfficiency in Australia, December 2002
817 V. Anton Muscatelli, Patrizio Tirelli, and Carmine Trecroci, Monetary and Fiscal PolicyInteractions over the Cycle: Some Empirical Evidence, December 2002
818 Claude Hillinger, A General Theory of Price and Quantity Aggregation and WelfareMeasurement, December 2002
819 Erkki Koskela and Ronnie Schöb, Optimal Capital Taxation in Economies withUnionised and Competitive Labour Markets, December 2002
820 Sheilagh Ogilvie, Guilds, Efficiency, and Social Capital: Evidence from German Proto-Industry, December 2002
821 Hans Gersbach and Verena Liessem, Financing Democracy, December 2002
822 Costas Hadjiyiannis, Panos Hatzipanayotou, and Michael S. Michael, Optimal TaxPolicies with Private-Public Clean-Up, Cross-Border Pollution and Capital Mobility,December 2002
823 François Ortalo-Magné and Sven Rady, Homeownership: Low Household Mobility,Volatile Housing Prices, High Income Dispersion, December 2002
824 Syed M. Ahsan and Panagiotis Tsigaris, Measuring the Social Discount Rate underUncertainty: A Methodology and Application, December 2002
825 Kai A. Konrad, Altruism and Envy in Contests: An Evolutionarily Stable Symbiosis,December 2002
826 Robert S. Chirinko and Huntley Schaller, A Revealed Preference Approach toUnderstanding Corporate Governance Problems: Evidence from Canada, December2002
827 Geir B. Asheim, Green National Accounting for Welfare and Sustainability: ATaxonomy of Assumptions and Results, December 2002
828 Andrea Gebauer, Chang Woon Nam, and Rüdiger Parsche, Lessons of the 1999Abolition of Intra-EU Duty Free Sales for Eastern European EU Candidates, December2002
829 Giacomo Corneo, Work and Television, December 2002
830 Vivek H. Dehejia and Yiagadeesen Samy, Trade and Labour Standards – Theory, NewEmpirical Evidence, and Policy Implications, December 2002
831 Geir B. Asheim and Wolfgang Buchholz, A General Approach to Welfare Measurementthrough National Income Accounting, December 2002
832 Aaron Tornell and Frank Westermann, The Credit Channel in Middle IncomeCountries, January 2003
833 Gebhard Flaig, Time Series Properties of the German Monthly Production Index,January 2003
834 Campbell Leith and Jim Malley, Estimated Open Economy New Keynesian PhillipsCurves for the G7, January 2003
835 Burkhard Heer and Bernd Süssmuth, Inflation and Wealth Distribution, January 2003
836 Erkki Koskela and Leopold von Thadden, Optimal Factor Taxation under WageBargaining – A Dynamic Perspective, January 2003
837 Carola Grün and Stephan Klasen, Growth, Income Distribution, and Well-Being:Comparisons across Space and Time, January 2003
838 Robert S. Chirinko and Ulf von Kalckreuth, On the German Monetary TransmissionMechanism: Interest Rate and Credit Channels for Investment Spending, January 2003
839 Sascha O. Becker, Andrea Ichino, and Giovanni Peri, How Large is the “Brain Drain”from Italy?”, January 2003
840 Albert Berry and John Serieux, All About the Giants: Probing the Influences on Growthand Income Inequality at the End of the 20th Century, January 2003
841 Robert Fenge and Martin Werding, Ageing and the Tax Implied in Public PensionSchemes: Simulations for Selected OECD Countries, January 2003
842 Robert Fenge and Martin Werding, Ageing and Fiscal Imbalances Across Generations:Concepts of Measurement, January 2003
843 Giovanni Andrea Cornia, The Impact of Liberalisation and Globalisation on IncomeInequality in Developing and Transitional Economies, January 2003
844 Peter Fredriksson and Per Johansson, Program Evaluation and Random Program Starts,January 2003
845 Bernd Hayo and Matthias Wrede, Fiscal Equalisation: Principles and an Application tothe European Union, January 2003
846 Syed M. Ahsan and Jaideep Oberoi, Inequality, Well-being and Institutions in LatinAmerica and the Caribbean, January 2003
847 Chang Woon Nam and Doina Maria Radulescu, The Role of Tax Depreciation forInvestment Decisions: A Comparison of European Transition Countries, January 2003
848 V. Bhaskar and Steinar Holden, Wage Differentiation via Subsidised General Training,January 2003
849 Paloma Lopez-Garcia, Labour Market Performance and Start-up Costs: OECDEvidence, January 2003
850 Christian Keuschnigg and Soren Bo Nielsen, Public Policy for Start-upEntrepreneurship with Venture Capital and Bank Finance, January 2003
851 Yin-Wong Cheung, Menzie D. Chinn, and Eiji Fujii, China, Hong Kong, and Taiwan: AQuantitative Assessment of Real and Financial Integration, January 2003
852 Gregory D. Hess, The Economic Welfare Cost of Conflict: An Empirical Assessment,February 2003
853 Douglas J. Cumming and Jeffrey G. MacIntosh, Comparative Venture CapitalGovernance. Private versus Labour Sponsored Venture Capital Funds, February 2003
854 Eckhard Janeba and John Douglas Wilson, Decentralization and International TaxCompetition, February 2003
855 Tapio Palokangas, Capital Accumulation and Employment Cycles in a Model ofCreative Destruction, February 2003
856 Brendan Walsh, When Unemployment Disappears: Ireland in the 1990s, February 2003
857 Luis H. R. Alvarez and Erkki Koskela, A General Approach to the Stochastic RotationProblem with Amenity Valuation, February 2003
858 Christian Schultz, Strategic Campaigns and Redistributive Politics, February 2003
859 Ernst Fehr and Joseph Henrich, Is Strong Reciprocity a Maladaptation? On theEvolutionary Foundations of Human Altruism, February 2003
860 Haizhou Huang, Dalia Marin, and Chenggang Xu, Financial Crisis, Economic Recoveryand Banking Development in Former Soviet Union Economies, February 2003
861 Pedro Cardoso and Bernard M.S. van Praag, How Sustainable Are Old-age Pensions ina Shrinking Population with Endogenous Labour Supply?, February 2003
862 Volker Meier, Efficient Transfer of Aging Provisions in Private Health Insurance,February 2003
863 Edward Castronova, Theory of the Avatar, February 2003
864 Robert S. Chirinko, Hans van Ees, Harry Garretsen, and Elmer Sterken, InvestorProtections and Concentrated Ownership: Assessing Corporate Control Mechanisms inthe Netherlands, February 2003
865 Bernard M.S. van Praag and Pedro Cardoso, The Mix Between Pay-as-you-go andFunded Pensions and what Demography has to do with it, February 2003
866 Ernst Fehr, Urs Fischbacher, Bernhard von Rosenbladt, Jürgen Schupp, and Gert G.Wagner, A Nation-Wide Laboratory. Examining Trust and Trustworthiness byIntegrating Behavioral Experiments into Representative Survey, February 2003
867 Frank Heinemann, The Inflationary Impact of Wage Indexation, February 2003
868 Eytan Sheshinski, Bounded Rationality and Socially Optimal Limits on Choice in aSelf-Selection Model, February 2003
869 M. Hashem Pesaran, Estimation and Inference in Large Heterogenous Panels with CrossSection Dependence, February 2003
870 Luis H. R. Alvarez and Erkki Koskela, On the Tree-Cutting Problem under Interest Rateand Forest Value Uncertainty, February 2003
871 Norbert Berthold and Rainer Fehn, Unemployment in Germany: Reasons andRemedies, February 2003
872 Clemens Fuest, Bernd Huber, and Philipp Tilleßen, Tax Policy and Entrepreneurship inthe Presence of Asymmetric Information in Capital Markets, February 2003
873 Eytan Sheshinski, Optimum and Risk-Class Pricing of Annuities, February 2003
874 Willi Leibfritz, Paul O’Brien and Jean-Christophe Dumont, Effects of Immigration onLabour Markets and Government Budgets – An Overview, February 2003
875 M. Hashem Pesaran and Allan Timmermann, How Costly is it to Ignore Breaks whenForecasting the Direction of a Time Series?, February 2003
876 Thorvaldur Gylfason and Gylfi Zoega, Education, Social Equality and EconomicGrowth: A View of the Landscape, February 2003
877 Robin Boadway and Jean-François Tremblay, Public Economics and StartupEntrepreneurs, February 2003
878 Erkki Koskela and Roope Uusitalo, The Un-Intended Convergence: How the FinnishUnemployment Reached the European Level, February 2003
879 Robert Fenge and Volker Meier, Pensions and Fertility Incentives, February 2003
880 Eytan Sheshinski, Note on Income Taxation and Occupational Choice, February 2003
881 A B Atkinson, Income Inequality in OECD Countries: Data and Explanations, February2003